whisper-v3-dyu4

This model is a fine-tuned version of openai/whisper-large-v3 on the abdouaziiz/wolof_lam_asr dataset. It achieves the following results on the evaluation set:

• Loss: 0.4723
• Wer: 0.3675

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 32000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8017	0.3106	500	0.7470	0.4962
0.6325	0.6211	1000	0.6371	0.3837
0.5929	0.9317	1500	0.5565	0.3274
0.3732	1.2422	2000	0.5384	0.4248
0.3639	1.5528	2500	0.5200	0.3033
0.3754	1.8634	3000	0.4936	0.4525
0.2005	2.1739	3500	0.4988	0.3129
0.1989	2.4845	4000	0.4852	0.4238
0.2155	2.7950	4500	0.4723	0.3675
0.1423	3.1056	5000	0.4926	0.2938
0.1136	3.4161	5500	0.5002	0.2962
0.1137	3.7267	6000	0.4891	0.2797

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3